6.1-stable review patch. If anyone has any objections, please let me know.
------------------
From: Alison Schofield alison.schofield@intel.com
[ Upstream commit b626070ffc14acca5b87a2aa5f581db98617584c ]
The compare function used to sort memblks into starting address order fails when the result of its u64 address subtraction gets truncated to an int upon return.
The impact of the bad sort is that memblks will be filled out incorrectly. Depending on the set of memblks, a user may see no errors at all but still have a bad fill, or see messages reporting a node overlap that leads to numa init failure:
[] node 0 [mem: ] overlaps with node 1 [mem: ] [] No NUMA configuration found
Replace with a comparison that can only result in: 1, 0, -1.
Fixes: 8f012db27c95 ("x86/numa: Introduce numa_fill_memblks()") Signed-off-by: Alison Schofield alison.schofield@intel.com Acked-by: Dave Hansen dave.hansen@linux.intel.com Reviewed-by: Dan Williams dan.j.williams@intel.com Link: https://lore.kernel.org/r/99dcb3ae87e04995e9f293f6158dc8fa0749a487.170508554... Signed-off-by: Dan Williams dan.j.williams@intel.com Signed-off-by: Sasha Levin sashal@kernel.org --- arch/x86/mm/numa.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/arch/x86/mm/numa.c b/arch/x86/mm/numa.c index e60c61b8bbc61..dae5c952735c7 100644 --- a/arch/x86/mm/numa.c +++ b/arch/x86/mm/numa.c @@ -961,7 +961,7 @@ static int __init cmp_memblk(const void *a, const void *b) const struct numa_memblk *ma = *(const struct numa_memblk **)a; const struct numa_memblk *mb = *(const struct numa_memblk **)b;
- return ma->start - mb->start; + return (ma->start > mb->start) - (ma->start < mb->start); }
static struct numa_memblk *numa_memblk_list[NR_NODE_MEMBLKS] __initdata;